Re: [win-pv-devel] v9.0.0 released

["Durrant, Paul" <pdurrant@xxxxxxxxxx>]

> -----Original Message-----
> From: Håkon Alstadheim <hakon@xxxxxxxxxxxxxxxxxx>
> Sent: 14 December 2019 13:34
> To: Durrant, Paul <pdurrant@xxxxxxxxxx>; Steffan Cline
> <steffan@xxxxxxxxx>; win-pv-devel@xxxxxxxxxxxxxxxxxxxx
> Subject: Re: [win-pv-devel] v9.0.0 released
> 
> Den 10.12.2019 22:12, skrev Håkon Alstadheim:
> >
> > Den 09.12.2019 22:20, skrev Steffan Cline:
> >> Does anyone have a specific config file they'd like to share that
> >> performs well with these newest drivers?
> >>
> >> On my Win 2016 Server, I removed the 8.2 drivers, rebooted, installed
> >> these and rebooted. It's been loading for about 30 minutes with the
> >> spinning dots non-stop.
> >>
> > Same here. Restored disk image from backup and tried again, and got
> > same second time. Left it for a day, still spinning. After destroying
> > it, at some point  I got "Can not access boot device" in the middle of
> > that pale blue screen. Trying to roll back using recovery point fails.
> > (That is the extent of the error
> 
> Den 11.12.2019 09:29, skrev Durrant, Paul:
> > Can someone post a log of the qemu output (with xen_platform_log in the
> events file) so we can see what the PV drivers are doing?
> >
> 
> New install of version 9 drivers. I've got some output now. Installed
> without removing old drivers this time, thinking that might work around
> the problem. No such luck.
> 
> After having killed my windows 10 domU and enabling xen_platform_log in
> qemu trace events, I got the below. Sorry I did not capture the first
> boot into the new drivers. I can do that for another run, but it takes
> some time to set up. Any other logging you would like to see?
> 
>    (more info after ".log ends")
> 

So your xenbus instance is coming up ok:

FFFF9B85D3779D00 (XP0001 XENBUS) [ACTIVE]

It is then creating VBD, VIF, IFACE, and CONS PDOs as expected. You have PDO 
revisions 0800000A, 0800000B and 09000000 - 09000007, so the bindings should be 
ok. Then I can see emulated disks and cd-rom coming up:

32056@1576324797.180483:xen_platform_log xen platform:
xenfilt|PdoCreate: FFFF9B85D3784040
(IDE\DiskQEMU_HARDDISK___________________________2.5+____\0.0.0)
32056@1576324797.180690:xen_platform_log xen platform:
xenfilt|PdoCreate: FFFF9B85D3784C50
(IDE\CdRomQEMU_QEMU_DVD-ROM_______________________2.5+____\0.1.0)
32056@1576324797.187770:xen_platform_log xen platform:
xenfilt|PdoCreate: FFFF9B85D37A7DA0
(IDE\DiskQEMU_HARDDISK___________________________2.5+____\1.0.0)
32056@1576324797.187960:xen_platform_log xen platform:
xenfilt|PdoCreate: FFFF9B85D302BD20
(IDE\DiskQEMU_HARDDISK___________________________2.5+____\1.1.0)

This is what would be expected as, to avoid a 0x7B BSOD, an upgrade of XENVBD 
must include an interim fall-back to emulated disk (since Windows no longer has 
a CDDB).

The fact that you don't get a 0x7B does imply that Windows is happy that it has 
a system disk and is instead waiting for something else to appear. We see no 
evidence of XENVBD loading (and it should, because the VBD PDO is present) so I 
guess I has to be something in the PnP subsystem that's wedging up. I've not 
seen anything like this before. I guess I should be able to repro by installing 
8.2.2 and then 9.0.0 over the top.

I notice you have a very big mmio_hole (3G) specified in your xl.cfg. I wonder 
if Windows is trying to allocate low memory and simply can't get enough. Could 
you try 'mmio_hole=2048' instead? Also, I notice you are on bleeding-edge Xen 
(4.13-rc)... Could you also try setting 'viridian=0' and see if that makes any 
difference?

  Thanks,

    Paul
_______________________________________________
win-pv-devel mailing list
win-pv-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/win-pv-devel